Re: [Dime] RFC 3588 - fixed positioned session-ID AVPs

Victor Fajardo [mailto:vf0213@gmail.com] writes:

Its good to have some clarity, 

Clarity is a wonderful thing & 3588bis could probably (still!) use a little
more of it but this point seems so blindingly obvious that further
wordsmithing would be just a waste of time.

though the statement below probably needs some re-org since SHOULD is
followed with a MUST and theres a conditional 'may' in between. Anyway, the
statement maybe redundant since each BNF already tells you where the
session-id can be found ..

my 2 cents.

On Wed, Sep 1, 2010 at 8:55 AM, David Lehmann <dlehmann@ulticom.com> wrote:

OK, BNF rules.    IMHO, this should be noted or clarified in section 8.8.
e.g.

"When present, the Session-Id SHOULD appear immediately following the
Diameter Header. Further, the message BNF may mandate that the Session-Id
MUST be positioned immediately following the message header.  Indeed, all
messages defined in this RFC require such a positioning. (see Section 3)"

--

David Lehmann

Ulticom, Inc.

856-787-2952

From: Victor Fajardo [mailto:vf0213@gmail.com] 
Sent: Tuesday, August 31, 2010 4:25 PM 

To: David Lehmann
Cc: Glen Zorn; dime@ietf.org
Subject: Re: [Dime] RFC 3588 - fixed positioned session-ID AVPs

yes. the BNF sets the positioning/sequencing rules. it's a common practice
in BNFs to place session-id near the head to optimize msg processing .. etc.

On Tue, Aug 31, 2010 at 3:22 PM, David Lehmann <dlehmann@ulticom.com> wrote:

So you are stating that the Diameter protocol itself does NOT require all
session-ID AVPs to follow immediately after the message header, but a
message's BNF may require it?  

--

David Lehmann

Ulticom, Inc.

856-787-2952

From: Victor Fajardo [mailto:vf0213@gmail.com] 
Sent: Tuesday, August 31, 2010 2:21 PM 

To: David Lehmann
Cc: Glen Zorn; dime@ietf.org
Subject: Re: [Dime] RFC 3588 - fixed positioned session-ID AVPs

On Tue, Aug 31, 2010 at 1:54 PM, David Lehmann <dlehmann@ulticom.com> wrote:

The existing text in 8.8 is contradicting the BNF.  

Which BNF though ? There maybe apps beyond 3588 that may not necessarily put
the session-id after the header in their BNF's. In that regard, the existing
text is not contradictory but is meant to be generalized specially because
it describes an AVP and is agnostic to any BNF.

my two cents,

victor

I am suggesting text that agrees and supports the BNF.  

If you don't want to modify the text to agree with the BNF, then I suggest
removing the existing text which contradicts it.

--

David Lehmann

Ulticom, Inc.

856-787-2952

From: Victor Fajardo [mailto:vf0213@gmail.com] 
Sent: Tuesday, August 31, 2010 11:52 AM
To: David Lehmann
Cc: Glen Zorn; dime@ietf.org
Subject: Re: [Dime] RFC 3588 - fixed positioned session-ID AVPs

Hi David,

On Tue, Aug 31, 2010 at 10:37 AM, David Lehmann <dlehmann@ulticom.com>
wrote:

Glen,

That is not what the spec says.  At least, the wording is not clear and,
IMHO, is misleading.  In which "some message" can the session-id AVP be in
any position?

It seems to me that the wording in section 8.8 should be:  "When present,
the Session-Id MUST appear immediately following the Diameter Header (see
Section 3)."

As Glen has mentioned, the BNF dictates the positioning of the AVP. If you
add this text you are adding new rules beyond the BNF.

regards,

victor

--

David Lehmann

Ulticom, Inc.

856-787-2952

From: Glen Zorn [mailto:gwz@net-zen.net] 
Sent: Tuesday, August 31, 2010 9:54 AM
To: David Lehmann
Cc: dime@ietf.org
Subject: RE: [Dime] RFC 3588 - fixed positioned session-ID AVPs

David Lehmann  <mailto:[mailto://dlehmann@ulticom.com%5d>
[mailto://dlehmann@ulticom.com] writes:

Hello,
Hello.
In RFC 3588 (and 3588bis), messages with session IDs are defined with the
session ID AVPs with a fixed position which is immediately following the
header. (e.g. section 8.3.1)  
Yes.
However, this is in contradiction with section 8.8 which states, "When
present, the Session-Id SHOULD appear immediately following the Diameter
Header (see Section 3)."
No.
By using "SHOULD", the spec is stating that the session-ID AVP could be in
any position in the message.

No, it is stating that the AVP could be in any position in some message.
The syntax of the existing messages in RFC 3588 is defined by the associated
BNF and in those messages the Session-Id AVP must immediately follow the
Diameter header.

Hope this helps.

 ~gwz

_______________________________________________
DiME mailing list
DiME@ietf.org
https://www.ietf.org/mailman/listinfo/dime